79 [This article was published in Working Papers in Athabaskan (Dene) Languages 2012. Alaska Native Language Center Working Papers, No. 11, eds. Sharon Hargus, Edward Vajda, Daniel Hieber. Fairbanks, AK: ANLC. 2013. Pp. 79-91. This version contains some corrections.] Vestigial possessive morphology in Na-Dene and Yeniseian1 Edward Vajda Western Washington University 1 Introduction External comparison with possessive constructions in the Yeniseian languages of Siberia suggests a diachronic explanation for morphological idiosyncrasies associated with Na-Dene possessed nouns, postpositions, directionals, and demonstrative prefixes. Section 2 discusses the nasal-class prefix that appears before certain inalienably possessed nouns in Athabaskan (Dene) languages. Section 3 intro- duces comparative Yeniseian morphology to propose that this element is a remnant of a generic posses- sive affix once regularly present between possessor and possessum in both families but surviving today in Athabaskan mostly before high frequency nouns. Section 4 considers Eyak, where, as is known, the l-qualifier is sometimes cognate with the Athabaskan nasal-class prefix (Krauss, in prep.). The compari- son with Yeniseian suggests that some instances of the Eyak d- and l-qualifiers may derive from fossil- ized possessive affixes, though most other qualifiers derive from anatomical nouns. Section 5 compares postpositional constructions in both families, which also show evidence of once having contained pos- sessive connectors. Section 6 considers directionals, defined by Leer (1989: 576) as “words that specify direction with regard to a frame of reference, such as a body of water”. Directionals in the two families have striking semantic and morphological parallels, including vestiges of possessive connectors. Sec- tion 7 examines evidence showing that Yeniseian and Na-Dene demonstrative prefixes were originally connected to the following stem by a possessive affix. Finally, section 8 considers non-canonical onset correspondences between Tlingit and Athabaskan-Eyak body-part nouns that may have arisen when the noun in Pre-Tlingit absorbed a prefix cognate to the nasal/lateral elements attested in Athabaskan- Eyak and Yeniseian possessive constructions. Section 9 summarizes these findings and considers a few unanswered questions brought to light by the discussion. 2 Nasal-class nouns in Athabaskan Possessive prefixes before certain inalienably possessed nouns in Dene (Athabaskan) languages involve a nasal element not present in conjunction with other nouns. Rice (1989: 211) provides the following 1. I am grateful to the volume’s co-editors, Sharon Hargus and Danny Hieber, for their helpful comments and questions, and for the opportunity to include the present article, which was not actually presented at the 2013 Athabaskan/Dene Conference. The conference was successful in great part thanks to Sharon’s professional expertize and interpersonal skills, which achieved a unique blend of historical and contemporary topics, along with a seamless integration of language revitalization with theoretical linguistics. 80 Working Papers in Athabaskan (Dene) Languages 2012 — Vajda forms, where the inalienably possessed noun –lá ‘hand’ requires nasal-class forms of possessive pre- fixes: (1) Slave noun with nasal-class prefix sįlá ‘my hand’ nįlá ‘your (sg.) hand’ bįlá ~ mįlá ‘his/her hand’ As is customary in Athabaskan linguistics, the hook below į in the forms listed in (1) indicates nasaliza- tion of the vowel. Non-nasalized allomorphs of Slave possessive prefixes – se- ‘my’, ne- ‘your (sg.)’, be- ~ me- ‘his/her’ – appear before other possessed nouns, often with an accompanying possessive suffix in the form of -é for alienable possession, or - ́ (high tone) for inalienably possessed nouns (Rice 1989: 215). Slave nouns that take the nasal-class prefix, like –lá ‘hand’, lack the possessive suffix. The examples of Slave alienably possessed nouns in (2) are cited from Rice (1989: 207), with the unpossessed noun pro- vided in parentheses: (2) Slave possessive constructions with non-nasal-class prefixes se-mbeh-é ‘my knife’ (mbeh ‘knife’) ne-tl’ul-é ‘your (sg.) rope’ (tl’uh ‘rope’) me-lį-é ‘his/her dog’ (lį ‘dog’) Body part nouns incorporated thematically into finite verb forms are not preceded by the nasal element, as in Slave k’etthíechu ‘s/he turns his/her head’ (Rice 1989: 647), where -tthí- ‘head’ is not pre- ceded by a possessive prefix and therefore lacks the nasal element as well. Homologous idiosyncrasies can be found elsewhere in Athabaskan in connection with posses- sive morphology and inalienably possessed nouns. Witsuwit’en, though lacking nasal vowels, contains several inalienably possessed nouns with an unexplained nasal-initial segment: –ntaq ‘forehead’, –nɣen ‘head’, –ntsəs ‘nose’, –ntl’at ‘temple’, –ntsən ‘arm, foreleg, sleeve’, –ntsətl ‘back of head’2. In Navajo, a high tone appears in possessive prefixes before some of the same inalienably possessed nouns that require nasal-class prefixes in Slave. The homologous nature of these elements can be illustrated by compar- ing the Navajo high-toned prefix in ší-la’ ‘my hand’ with the Slave nasal-class prefix in sįlá ‘my hand’, as contrasted with the regular prefixed Navajo ši-bééž ‘my knife’ and Slave se-mbeh-é ‘my knife’. Hoijer (1969: 157) reconstructed *ni- as the historical form of the Navajo high-tone element in pos- sessive prefixes based on internal reconstruction and comparison with cognate constructions meaning ‘(someone’s) hand/finger’ in other Athabaskan languages, including Gwich’in –ni-li’, Beaver –n-la’, and Tɬįchǫ (Dogrib)–n-la . Leer (1996, 2012: 1) reconstructs the Proto-Athabaskan nasal-class prefix as *n- rather than Hoijer’s *ni-. The survival of this element before cognate vocabulary in different geographic areas of the Atha- baskan world suggests an ancient, inherited pattern. The presence in Eyak of a cognate prefix with al- lomorphs -:n- ~ -la- also supports the archaic provenance of this feature of possessive constructions. The following examples, taken from Krauss’s discussion of Eyak nouns and qualifiers (Krauss, in prep.), show that the nasal allomorph predictably appears before coronals, while la- appears elsewhere: -ːn-daː’ ‘face’, 2. I thank Sharon Hargus for pointing out these forms. For more on Witsuwit’en possessive morphology, see Hargus (2007: 229–232, 618-620). Working Papers in Athabaskan (Dene) Languages 2012 — Vajda 81 -ːn-dalah ‘antler, horn’, -ːn-ch’it’ ‘forehead’, -lɑ-ɢaːnš ‘part of face below nose’, -la-qah ‘head’, -lɑ-quhɬ ‘cheek’, -la-χu’ ‘facial hair’, -la-wahsq’ ‘temple. In these particular examples, the alternating qualifier forms -ːn- ~ -la- apparently reflect the Proto-Na-Dene nominal root *-ñan’, meaning ‘face’ (Leer 2012: 1). In other cases, such as tsaː-la-q’aχ ‘jellyfish’ (< tsaː ‘rock’ + q’aχ ‘fat’) and tsaː-la-χaɬ ‘gravel on beach’ (< tsaː ‘rock’ + χaɬ ‘granular substance?’), the Eyak l-qualifier cannot be etymologized as deriving from an anatomical noun. The next section introduces Yeniseian comparanda to argue that the Athabaskan nasal-class prefix and some (but not all) instances of the Eyak l-qualifier are vestiges of an ancient ge- neric possessive affix. 3 Yeniseian possessive morphology Yeniseian is a family of several languages once spoken across much of central and southern Siberia, but now represented solely by Ket, which has fewer than fifty elderly speakers. The family once contained at least two primary branches – Ket and Kott – and has been hypothesized to be genealogically related to Na-Dene on the basis of shared core vocabulary and morphology (Vajda 2010). Like Na-Dene languag- es, Yeniseian uses preposed pronominal markers to express possession. However, there appears to be no trace of any possessive suffix following the possessed noun. Nor is there a distinction between inalien- able and alienable possession. Body part nouns and kinship terms in the three Modern Ket dialects (Northern, Southern, and Central Ket) may be uttered either with or without a possessor indicated. Examples of Ket nouns in possessive constructions appear in (3): (3) Northern Ket possessed nouns3 b-ki’s ‘my leg’ k-tɨ’ ‘your (sg.) head’ d-go’d ‘her/its rump’ da-qāˑre ‘his fur’4 na-qo’n ‘our/your/their conifer needles’ The possessive markers in Modern Ket are actually special clitics. In fast speech, they normally attach to any available preceding word. When pronounced sentence initially or preceded by a pause they pro- cliticize to the following possessum noun. The example in (4) is adapted from the discussion of South- ern Ket possessive constructions in Vajda (2008: 188-190): (4) sul-d ètl blood-3INAN.POSS color ‘the color of blood’ The variable phonological behavior of Ket possessive morphemes has been attributed to a typological shift in favor of root-initial phonological words, which developed under the influence of the surround- ing suffixing languages (Vajda 2009: 486-488). The so-called “genitive case” of Ket nouns and pronouns is, in fact, simply the possessive morpheme encliticized to the preceding possessor noun in fast speech. 3. Ket phonemic prosody is transcribed as: v̄ ˑ for high-even tone on a half-long vowel; v’ for abrupt rising tone ending in creaky voice or full glottal closure; vː (or vv) for rising-falling tone on a geminate vowel, and v̀ for falling tone. The symbol /ə/ repre- sents a back rather than central unrounded vowel, which is allophonically realized as mid-high [ɤ] under high-even tone and as mid-low [ʌ] elsewhere. The symbol